Why D3 engine?

[Gabrobot]

FireFly said:
Future game engines exceed Doom 3. Current game engines are on par with Doom 3, feature-wise. What does the Doom 3 engine do that other engines don't?

I mean, the sound system is pretty simple, the specularity system is incredibly limited, there's no support for pre-computed shadowing, there's no support for post process effects, there's no advanced squad based A.I system.

The sound system may be simple, but the reason for that was because it was easier for the developers to use and get the results they wanted (the sound engine that was in Doom 3 before was very advanced and had sound physics that could bounce sounds off walls and all that shit). If developers want an advanced sound system they can either use the old one (they have access to such things even if it never made it into the final Doom 3 engine used in Doom 3) or they can use the EAX one that'll soon be released in the 1.3 patch. I'd say D3 has no problems in the sound department.

Specularity system is limited? You mean there's no gloss map support? All one needs to do to add support for gloss maps is add some code in the interaction shader...mods have already done this, it's certainly no problem for developers.

No support for precomputed lighting...why the hell would you want precomputed lighting? Even UE3 doesn't use precomputed lighting even though it still uses precomputed shadows. Real time lighting is far more realistic looking and a lot more flexible. Did you mean precomputed shadows? Precomputed shadows take tons of time to compute (offline) but still often end up low res and often glitchy. They make doing large areas very hard because of the resolution you need on the lightmap in order for the shadows to not look like crap (blocky mess)...this takes a long time to compute, and it takes up tons of memory when running the level.

No post processing effects? You can run shaders on the screen...what kind of post-processing effects did you have in mind?

How do you know there's no advanced AI system? You obviously didn't do any research since Doom 3 has an extremely powerful AI system powered by the engines scripting system. Doom 3 hardly even scratched the surface...I presume Quake IV is going take better advantage of it from what the articles have said.

And as far as UE3, for the most part what has wowed people is the artwork...you can get just about the same thing with the Doom 3 engine since most of it is simply geomtry with high-res normal maps. You'd need to add HDR to D3, but that's not to hard a thing to do (Doom 3 already has some HDR code...the biggest problem with adding HDR is actually just implemmenting controls for it into all the different tools). UE3 uses soft shadows on a few things, but most of the real-time shadows are actually stencil shadows like Doom 3 has. Of course if you talk about the tools UE3 has, then there's no contest...UE3 far surpasses everything we currently have.

[Drazula]

You're proving his point. To take full advantage of the tech requires focus on gameplay, i.e examining "the potential of area they are in".

You are not saying the same thing he is. He said Doom 3 was not a good tech demo because it didn't take full advantage of the engine. In other words, a good tech demo takes full advantage of the engine. You are saying good gameplay takes full advantage of an engine.

He calls full advantage a tech demo, you call it gameplay. Which is it? And no, it can't be both since a "tech demo" implies no gameplay.

BTW, what I said was that none of id's games took advantage of the engine, therefore they were not good "tech demos". They were great games.

Future game engines exceed Doom 3. Current game engines are on par with Doom 3, feature-wise. What does the Doom 3 engine do that other engines don't?

You'll have to wait until I stop laughing for me to reply sensibly to this... it may take a while. Features are only as good as their implementation. By your logic, the Serious Engine is better than all of them because it has more features. It is how good those features are that matters.

I mean, the sound system is pretty simple, the specularity system is incredibly limited

Sound is simple but the results are good. As for the specular system, that is your opinion.

there's no support for pre-computed shadowing,

Uh, that's a good thing. All lighting is treated the same way. That's the point.

there's no advanced squad based A.I system.

You mean that POS static node system used in Source? The one I had to gut out so I could put a real AI system in place? I would have been better off with Doom 3. At least it is not pretentious about AI. (Far Cry's AI is still the best.)

[Kristian Joensen]

Kristian Joensen said:
What physics engine does Prey use ?

Oh and btw is there any code at all in there from the old Prey ?

[Krid]

Kristian Joensen said:

Oh and btw is there any code at all in there from the old Prey ?

What I want to know is how much of the nifty stuff shown in these videos is going to be in the game:

http://www.imhz.com/videos/Prey_E3_1998_IMHz.avi
http://atlas.secs.oakland.edu/~lcmat..._1998_Demo.mpg

I'm mostly intrigued by the portal system, as AFAIK it hasn't been used since the days of Build, and then only a fairly crude implimentation.

[Kristian Joensen]

Portals are in.

[Kev_Hectic]

Nevermind, looks like Joe already took all the links down.

[FireFly]

Drazula said:
You're proving his point. To take full advantage of the tech requires focus on gameplay, i.e examining "the potential of area they are in".
You are not saying the same thing he is. He said Doom 3 was not a good tech demo because it didn't take full advantage of the engine. In other words, a good tech demo takes full advantage of the engine. You are saying good gameplay takes full advantage of an engine.

He calls full advantage a tech demo, you call it gameplay. Which is it? And no, it can't be both since a "tech demo" implies no gameplay.

Both

A game demo can simply be an effective demonstration of the tech. He gives the example of HL2, a game which relies on its game design to show the full power of its tech. And again, take his physics systems example. In Doom 3 the physics system is fully present. According to your definition Doom 3 should be a great tech demo, however it's not because the applications of the tech have barely been touched upon.

So id aren't utilising their engine properly because they're concentrating more on the tech itself, which in turn makes that tech less marketable. This is why the "but they didn't utilise the tech properly" argument is no defence.

You'll have to wait until I stop laughing for me to reply sensibly to this... it may take a while. Features are only as good as their implementation. By your logic, the Serious Engine is better than all of them because it has more features. It is how good those features are that matters.

What makes you think feature-wise simply refers to the sheer number of features?

Sound is simple but the results are good.

Why settle for good when you can have great!

As for the specular system, that is your opinion.

You're right. It also happens to be Carmack's opinion.

"human characters have a kind of plastic look about them. This is because there's only a simple specular highlight system in place -- that's the part of the engine that renders light hitting a surface. DOOM 3's basic system looks great on surfaces that are dull or plastic, but it's not good at rendering high-gloss surfaces such as polished, shining steel."

http://pc.gamespy.com/pc/doom-3/539049p1.html

"What we call specular maps in Doom 3 are more commonly called "gloss maps," where it's just affecting the
intensity of the specular highlight, but we now also add in new technology, the ability to change the breadth of the specular highlight. That lets you do a lot of interesting things with... the highlight that we've got in Doom is
really quite broad for a specular highlight, and it's about what you'd get on a really dull plastic; something that wasn't very shiny, it's a kind of fairly broad, spread out thing. You don't get anything that looks like a really good metallic highlight, or things that would be shiny cast plastic, so there's a lot of neat stuff that you get just
playing with that, and going ahead and having some that are even broader and some that tighten down a whole lot to give you bright little pin-point highlights on there."

http://www.gamedev.net/community/for...opic_id=266373

Uh, that's a good thing. All lighting is treated the same way. That's the point.

Why do you want to treat all light in the same way? You're taking a mathmatical principle over visual judgement.

You get a massive boost in performance combined with a boost in lighting detail. And of course you can tweak the lighting models until the difference becomes nearly imperceptible.

You mean that POS static node system used in Source? The one I had to gut out so I could put a real AI system in place? I would have been better off with Doom 3. At least it is not pretentious about AI. (Far Cry's AI is still the best.)

Some squad combat is better than no squad combat at all. But why bring up Source? What about the Far Cry and F.E.A.R. engines?

In response to Gabrobot (I had a response but my browser crashed):

The sound system may be limited because id fell foul of excessive sophistication, but that doesn't invalidate the entire paradign. I didn't realise all previous code was available for easy integration.

I actually said "advanced squad based A.I", so I was talking about the group combat behaviour (working together, flanking, laying down fire etc.) And again, I said pre-computed shadowing so I wasn't talking about raytracing or anything. Doing precomputed shadowing should only yield a performance hit while the lighting is being calculated. After that it's baked into the level. Also texture size shouldn't be a problem because it's being rendered offline (and Carmack is already doing this stuff in real time anyway).

UE3 actually uses a lot of soft shadows. Stencil shadows are only utilised for moving lights. All shadows cast by static lights are soft. I agree that art is a lot of the appeal but that doesn't mean there isn't a hug gape technology wise between the two engines.

[Gabrobot]

FireFly said:
So id aren't utilising their engine properly because they're concentrating more on the tech itself, which in turn makes that tech less marketable. This is why the "but they didn't utilise the tech properly" argument is no defence.

This would be a good argument except that D3's tech was already pretty much done in 2003. Carmack talked about how much time he had to kill which he spent polishing the engine (he said he polished it more than any engine he's worked on before) and experimenting with technology for his next engine (which is where the HDR and shadow buffer cvars came from). Also, there are a couple test levels with two vehicles, with models and everything. It wouldn't have been hard to polish them and put them in the game, but id didn't because they didn't think it fit the design of the game they wanted.

FireFly said:Why do you want to treat all light in the same way? You're taking a mathmatical principle over visual judgement.

You get a massive boost in performance combined with a boost in lighting detail. And of course you can tweak the lighting models until the difference becomes nearly imperceptible.

Workflow. This is especially important today with the incredible amounts of detail artists have to create. Pre-computing things throws a wrench into the workflow. I have to work with lightmaps in my Jedi Academy levels and it's a real pain in the arse to have to compile a level, look at the lighting, go back into the editor and tweak it, compile again, rinse repeat. In Doom 3 the artist can render the level in real-time and tweak things to perfection in a tiny fraction of the time it would otherwise take. What you get is better quality much quicker (which means even more good stuff). Sure it runs slower, but modern computer have no problem running it. If an engine is running at 120fps on current computers, it doesn't mean it's well designed, it means it's out of date.
(Referring to the Source engine there, btw)

FireFly said:I actually said "advanced squad based A.I", so I was talking about the group combat behaviour (working together, flanking, laying down fire etc.)

That's entirely up to the programmer who writes the AI scripts...the point is the AI scripting system is powerful enough so that you can actually do that. TinMan made a squad combat mod, and Quake IV promises complex squad AI that flanks, distracts, ect.

FireFly said:And again, I said pre-computed shadowing so I wasn't talking about raytracing or anything. Doing precomputed shadowing should only yield a performance hit while the lighting is being calculated. After that it's baked into the level. Also texture size shouldn't be a problem because it's being rendered offline (and Carmack is already doing this stuff in real time anyway).

What I was referring to is something I've had to deal with in my Jedi Academy levels. In large levels with terrain, the lightmap data becomes huge (real-time shadowing only needs to process what's visible...when a lightmapped level is loaded, it needs to load the whole lightmap into memory). Early Q3 levels with terrain (such as those in Team Arena) didn't have lightmaps at all. With games using a diffuse map, normal map, specular map, and possibly a gloss and luminosity map, then a ton of lightmap data takes a large toll on the quality of all those other maps. UE3 isn't quite as bad here because it only has shadow masks (it does all lighting in real-time) but it still isn't as elegant as a unified system. Also, there are bound to be glitches and such in an engine with several shadowing methods used together...with a single method, the system can be polished to a much greater degree.

And again, workflow is a huge thing to consider. I think you have to be a level editor to really appreciate what real-time lighting let's you do.

Anyway, I'm done ranting for now.

[FireFly]

Gabrobot said:
This would be a good argument except that D3's tech was already pretty much done in 2003. Carmack talked about how much time he had to kill which he spent polishing the engine (he said he polished it more than any engine he's worked on before) and experimenting with technology for his next engine (which is where the HDR and shadow buffer cvars came from). Also, there are a couple test levels with two vehicles, with models and everything. It wouldn't have been hard to polish them and put them in the game, but id didn't because they didn't think it fit the design of the game they wanted.

And why is the game design limited? Because they are tech rather than gameplay focused!

If they wanted they could have made full use of the physics system, it was one of the things they talked about a lot in early previews. They chose to downplay its role.

Workflow. This is especially important today with the incredible amounts of detail artists have to create. Pre-computing things throws a wrench into the workflow. I have to work with lightmaps in my Jedi Academy levels and it's a real pain in the arse to have to compile a level, look at the lighting, go back into the editor and tweak it, compile again, rinse repeat. In Doom 3 the artist can render the level in real-time and tweak things to perfection in a tiny fraction of the time it would otherwise take. What you get is better quality much quicker (which means even more good stuff).

As I understand it though, workflow isn't smooth anyway. Level developers are waiting on tech, on programmer assistance and other art assets. So it's not just a case of "ok, we've done the lighting quicker so we'll automatically complete the level quicker".

I'd like to see some real world evidence as to the effect this sort of WYSIWYG editing has, especially in comparison to the quickest pre-computed solutions.

Sure it runs slower, but modern computer have no problem running it. If an engine is running at 120fps on current computers, it doesn't mean it's well designed, it means it's out of date.
(Referring to the Source engine there, btw)

Performance gained from using a pre-computed lighting model can be used on other areas, like texture or model 'resolution'. And of course with lower overhead, owners of lesser graphics cards don't have to sacrifice as much to get playable frame rates.

That's entirely up to the programmer who writes the AI scripts...the point is the AI scripting system is powerful enough so that you can actually do that. TinMan made a squad combat mod, and Quake IV promises complex squad AI that flanks, distracts, ect.

Ok, but how big a step up is that from writing everything yourself?

UE3 isn't quite as bad here because it only has shadow masks (it does all lighting in real-time) but it still isn't as elegant as a unified system. Also, there are bound to be glitches and such in an engine with several shadowing methods used together...with a single method, the system can be polished to a much greater degree.

What sort of glitches?

And again, workflow is a huge thing to consider. I think you have to be a level editor to really appreciate what real-time lighting let's you do.

I understand that, but I'm speaking as a gamer who wants the maximum performance and eye candy possible. Valve managed very effective lighting design despite a hugely time consuming map compile system, so why can't the rest of the world? And shadow masks should have a much lesser compile time anyway.

[ADM]

Lengis said:
Quake 4 will be fine, Prey will be fine, but they both still look exactly like Doom 3.

See this is proof that you just bullshit all the time Lengis. Ok we've seen Quake 4 screens but have you seen the new Prey screens?

No? So how do you know it'll look exactly like Doom 3?

An engine is an engine.. just a bunch of code, it's up to the Art team in the end to make it look different then Doom 3. Just because it's the same engine doesn't mean they have to look remotely similar.

Plus who even said they arn't going to make changes to the rendering engine?

[John]

Um...I've seen the Prey shots, and I'll go on a limb and say they look like Doom 3.

[Kev_Hectic]

^ Yeah, the whole PC Gamer article was posted on the board yesterday.





<font color="#1F1F22">Move along, nothing to see here...</font>

[ADM]

Well there you have it

I apologise.

[Drazula]

I thought the style of the creatures seemed more Unreal-like than Doom 3.

[Cerberus_e]

as long the AI is good like the review says (you never know: think of doom 3 and HL2)
because I'm tired of shooting seek-and-destroy-AI type monsters in games that only throw few enemies at a time at you.
doom 3 and half-life 2 are good but they only throw 1-2-3 enemies at you at the same time, and they have seek-and-destroy AI (HL2: except on hard, then the combine has hide-AI, but still no variaton, and it's only for the combine and is stille easy), which makes it less good.
painkiller and doom 2 also have seek-and-destroy AI monsters, but they come in groups of 20-30-40, which makes it fun.
I didn't mind it in doom 3 because the game is unique to some extent, but quake 4 and prey look like doom 3 so I don't want the same.
another exception to this rule would be far cry, because:
1) there are tons of other INTELLIGENT enemies
2) it only happens in a certain part of the game (the middle part)
3) the big/open environment makes it all good.

[Imfamous]

Drazula said:
I thought the style of the creatures seemed more Unreal-like than Doom 3.

[ADM]

Drazula said:
I thought the style of the creatures seemed more Unreal-like than Doom 3.

True.

It's definitely got it's Doom 3 qualities but all in all I think it's quite impressive and a far cry (heh) from Doom 3's dark style.

[Lengis]

John said:
Um...I've seen the Prey shots, and I'll go on a limb and say they look like Doom 3.

Nuff said^

[FireFly]

Drazula said:
I thought the style of the creatures seemed more Unreal-like than Doom 3.

The art design does, but there's something about the quality of the skin and lightiing that make the monsters seem very Doom 3-like.

It's like a version of Doom 3 with truly twisted enemies. Those Hounds look nasty.

[John]

It looks like a really badass Doom 3 expansion.

[Ivan]

John said:
It looks like a really badass Doom 3 expansion.

It looks badass, but I still have a problem with the too-close-similitute with D3.

[Lengis]

It looks better than Doom 3, I think we can agree on that at least.

[Cerberus_e]

looking better? looking the same? looking like far cry?
what strange opinions will I yet hear

[KaiserSoze]

graphics a game do not make.

gameplay guys, gameplay!

[FireFly]

Graphics + gameplay = game. Both are important.

[KaiserSoze]

i will take fun gameplay over graphics any day of the week.

[Lengis]

KaiserSoze said:
graphics a game do not make.

gameplay guys, gameplay!

You aren't telling us anything we don't know. But graphics are supposed to help serve the gameplay, and the atmosphere. If prey was Cell Shaded, it would be a completely different game. Humans are visual animals, and we react accordingly.

[Duoae]

KaiserSoze said:
i will take fun gameplay over graphics any day of the week.

Kaiser, you say that so often these days i think that it might save you some writing time if you just put it in your sig!

[FireFly]

KaiserSoze said:
i will take fun gameplay over graphics any day of the week.

Of course, but imagine fun gameplay with perfect graphics. That's why graphics are so important, they're the final layers of icing that give that deeper flavour.

Do you want Prey to just provide an enjoyable game, or would you like an enjoyable world and experience?

[John]

I definatley agree that the game is better looking than Doom 3. The picture with the lights on the floor looks simply amazing. And I am definatley hoping to see some outdoor elements;

seeing as how this game isn't based on Mars, so it doesn't have an excuse to look bland.

[Cerberus_e]

it's on a space ship instead, even less outdoor environments than mars

[John]

Nooooooo!

[FireFly]

But that's what's so cool. The ship is its own outdoors. And you get your own space pod \0/

[Kev_Hectic]

^ It might not be limited to just spaceship environments though. In the preview it says that there will be some vehicles to drive around in, including one that's a small space ship. If you look carefully at the screenshots, you can see that the player appears to be flying around in a small spacecraft on the moon.

EDIT: beaten by firefly!

[Orochi Avlis]

Some things in the environment look like they came directly from Doom 3. Or is it just me?

[Gabrobot]

FireFly said:
And why is the game design limited? Because they are tech rather than gameplay focused!

If they wanted they could have made full use of the physics system, it was one of the things they talked about a lot in early previews. They chose to downplay its role.

I'd like to point out that there was a lot of bad feeling at id Software with the decision to do a Doom game (a couple pro-Doom people were even fired by the lead artists). It's interesting to note that before that, id Software was actually looking at doing a fantasy themed RPG (they had even started design work on it). Carmack is definitely tech focused, but he knows that and lets the artists do the actual design. Things were omitted because their goal was not about showing off their tech. Doom 3 barely even scratched the surface of the engine's power, even considering computer hardware.

FireFly said:As I understand it though, workflow isn't smooth anyway. Level developers are waiting on tech, on programmer assistance and other art assets. So it's not just a case of "ok, we've done the lighting quicker so we'll automatically complete the level quicker".

I'd like to see some real world evidence as to the effect this sort of WYSIWYG editing has, especially in comparison to the quickest pre-computed solutions.

I think you misunderstood what I meant about the advantages of getting stuff done faster. If you can do something as time consuming as lighting much faster, than you can concentrate your efforts on things such as interactivity, less linear design, larger more detailed levels, ect. Even if nothing else, it means you can tweak the flow and gameplay of your level more instead of spending it tweaking lighting. If things are far enough along for the level designer to be making a level, they probably won't be held back from doing a lot of these things because of other factors.

As for real world evidence, I can't really say anything about that, but I know there is a level designer for Prey around here...perhaps he might be able to better compare workflow.

FireFly said:Performance gained from using a pre-computed lighting model can be used on other areas, like texture or model 'resolution'. And of course with lower overhead, owners of lesser graphics cards don't have to sacrifice as much to get playable frame rates.

You're talking about different things here...doing shadows in real-time doesn't take up memory which would normally be used for textures (that's what lightmaps do). The reason Doom 3 has lower resolution textures is simply because it uses more of them (every material has several layers). And while it's true that older video cards have to turn settings down, it still looks great. On my older computer (P3 866Mhz, Geforce 4) it still looks pretty damn good. (Although the CPU makes it run a bit slow...the outdoor levels actually run the fastest, going about 40FPS when a lot of indoor levels run at 5-10FPS in the worst parts).

FireFly said:Ok, but how big a step up is that from writing everything yourself?

It enables you to work at a higher level and concentrate more on the AI since you don't have to worry about low level stuff.

FireFly said:What sort of glitches?

Overhead from using several methods may make it run slower than if it had a single well optimized system. It's liable to run into problems with lots of lights (Far Cry for example has a cap on the number of dynamic lights that can cast shadows...that's why the flashlight doesn't cast shadows) due to all the interaction between different systems. I obviously can't speak directly about UE3 (it may be it works fine), but I still think a unified system is more efficient and dependable.

FireFly said:I understand that, but I'm speaking as a gamer who wants the maximum performance and eye candy possible. Valve managed very effective lighting design despite a hugely time consuming map compile system, so why can't the rest of the world? And shadow masks should have a much lesser compile time anyway.

It took Valve six years to make HL2, though.

Also shadow masks still take several hours on top of the line systems (lighting isn't what makes compiling lightmaps take so long...it's the shadow stuff).

[FireFly]

Gabrobot said:
I'd like to point out that there was a lot of bad feeling at id Software with the decision to do a Doom game (a couple pro-Doom people were even fired by the lead artists). It's interesting to note that before that, id Software was actually looking at doing a fantasy themed RPG (they had even started design work on it). Carmack is definitely tech focused, but he knows that and lets the artists do the actual design. Things were omitted because their goal was not about showing off their tech. Doom 3 barely even scratched the surface of the engine's power, even considering computer hardware.

I agree, but the fact remains that their direction is very tech focused, and is geared towards not fully exploiting the gameplay ramifications of that technology, which is why many of their games are branded tech demos.

As for real world evidence, I can't really say anything about that, but I know there is a level designer for Prey around here...perhaps he might be able to better compare workflow.

Yes, I'm still not sure what the qualitative benefits are, and I'd like to see some insight.

You're talking about different things here...doing shadows in real-time doesn't take up memory which would normally be used for textures (that's what lightmaps do).

I know but I'm talking about sheer performance. Although actually real time-shadows for Carmack's next engine will be created using shadow maps.

What I'm saying is, couldn't you use that extra power to add more detail to the environment, or even reduce the hit of other effects like HDR?

It took Valve six years to make HL2, though.

Only two years level design though. How many years of level design for Doom 3

Also shadow masks still take several hours on top of the line systems (lighting isn't what makes compiling lightmaps take so long...it's the shadow stuff).

Ok, thanks.

Edit: In UE3, how are the shadow masks generated? I thought maybe they were just using a simple shadow mapping system.

I mean Carmack is going to be using shadow maps for all the real time lighting, so why not just bake all the static shadows into the map using this system? Each shadow could be recalculated when an object is moved in the editor.

[pegisys]

I know but I'm talking about sheer performance. Although actually real time-shadows for Carmack's next engine will be created using shadow maps.

What I'm saying is, couldn't you use that extra power to add more detail to the environment, or even reduce the hit of other effects like HDR?

I thought it was the shadow buffer instead of the zbuffer, and from what I heard he is using both, a shadow buffer when light is going in one direction and the zbuffer for multiple shadows from one object from multiple lights, I don't know how true that is but it's what I heard

and I thought a shadow maps are static so you can do anything realtime with them

KaiserSoze said:
graphics a game do not make.

gameplay guys, gameplay!

Your right, but graphics that make full use of the hardware available do increase the wow factor of the game significantly I think.

[supermeerkat]

What little has been shown of Prey appears to look like D3. Howsabout waiting for the released game, then playing it and then making comments about the style of the game? Or am I missing something?